Jan Wassenberg

Yes, it is an unfortunate reality that current systems are imbalanced enough to strand a large fraction of cores for many workloads (exception: branchy code, which is still quite common)....

Hi, this can safely be ignored, but we do fix it in https://github.com/google/highway/commit/51d5f80987e4dc4771a9f12ccbb548e9416503e6. Is it still happening even after that change? The background is: we're checking that the CPU supports...

Hi, fix 9b6b473 is helpful, but it only resolves a warning. The results should be correct both before and after the fix.

Hi @kleisauke , thanks for following up. Looks like I was mistaken, this line `bits &= ~min_avx2;` seems not to promote min_avx2 to 64-bit. Thus the upper half of the...

Hi @nielskm, sounds good, we'd welcome such a patch. For the general approach, SO has some discussion (see also links): https://stackoverflow.com/questions/34066228/how-to-perform-uint32-float-conversion-with-sse To understand what each intrinsic is doing, we can...

@nielskm wow, congratulations, this is impressive work - awesome that you've also managed to update RVV/SVE! I've left a few comments which are mostly minor style/naming. Yes, our internal pipeline...

@nielskm Nice, tests pass. Getting cross-platform SIMD in just one test-edit iteration is solid work :) Just waiting for an additional internal review now, then this will go in.

:) I can confirm that HEAD compiles with gcc 11 on aarch64, as mentioned in g3doc/release_testing_process.md. The last four targets to be built are memory_test, sort_test, dot_test, logical_test. Looks like...

I've narrowed it down to the tests for LoadInterleaved3 or 4 - either causes long builds. Without those, aarch64 gcc 12 builds of memory_test take about 18s on my machine....

:) FYI filed https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106041.